[time-nuts] A silly question ...

Tom Van Baak tvb at LeapSecond.com
Mon Oct 1 04:38:01 EDT 2018


> Dave B. (G0WBX)
> 
> PS: I do like the idea of setting up a camera to take a photo of the
> 'scope every hour or so! Not practical for many I guess, but it
> illustrates the point well. But the aliasing opportunity I think would
> be perhaps too great, in essence being a sampled data system by then. 
> Also, one then needs an accurate 1 hour timer! And so it goes on ;-)

For hourly photos consider using a webcam or smart phone & time-lapse photo app.

Here's an example of measuring time drift in mains with a photo every 15 minutes. Watch how the red seconds hand drifts by up to 4 seconds during the day: http://leapsecond.com/pages/tec/mains-clock-ani.gif

About the aliasing, yes, one must keep that in mind. If the drift between your 10 MHz signals is erratic or if you slip by more than half of 100 ns per hour there could be ambiguity in your interpretation of the photos. That aliasing must be avoided.

There are two easy solutions:
1) Take a photo more often than once per hour. For example, once a minute improves the possibility of ambiguity by a factor of 60.
2) Divide the frequency from 10 MHz to 100 kHz. Now your cycles are 10 us instead of 100 ns, which reduces the chance of undetected cycle slip by a factor of 100.

Taking this to an extreme, imagine taking a photo every second, and imagine dividing your 10 MHz down to 1 Hz. Now you've reduced the chances of aliasing by tens of billions. And ... now you know why almost all timing measurements are done with a TIC as the "camera" and 1PPS as the "cycle".

This is also why, since 1972, UTC does time jumps (aka leap seconds) in steps of exactly 1 second instead of any smaller value -- it perfectly fools all the TIC's in the world.

/tvb




More information about the time-nuts mailing list